pm2 cluster mode uses max cpus

[arelra]

What does this change?

Following #8789 we can try running pm2 in cluster mode.

This PR sets the number of worker processes equal to the max number of cpus available.

This would increase the available cpus from 1 to 2 for our prod instance size of t4g.small.

Why?

Improve request latency and cpu utilisation of our instances.

Before

dotcom-+    1366  17:08   1:07 node /home/dotcom-rendering/rendering/dist/server.js

After

dotcom-+    1404  14:47   0:04 node /home/dotcom-rendering/rendering/dist/server.js
dotcom-+    1413  14:47   0:08 node /home/dotcom-rendering/rendering/dist/server.js

[github-actions]

Size Change: 0 B 🆕

Total Size: 0 B

_{compressed-size-action}

[georgeblahblah]

So exciting!

From a PR perspective, I think you might need to rebase against main to bring in the changes which make the AR CI build job for the required status check run.

[AshCorr]

I'm not totally convinced on PM2 yet, I find its lack of response to the AGPL question to be quite alarming and would prefer to find an alternative "process manager" if possible. But since we are using PM2 it seems a waste to not make use of our extra compute power! Good PR!

[arelra]

I'm not totally convinced on PM2 yet, I find its lack of response to the AGPL question to be quite alarming and would prefer to find an alternative "process manager" if possible. But since we are using PM2 it seems a waste to not make use of our extra compute power! Good PR!

I agree we should look for alternatives. Although pm2 is very mature the infrequent updates are concerning too.

Have suggested some next steps here.

In the mean time given cluster mode is easy to switch on, it will be interesting to see how it performs for us.